EXAFS setup is shown in Figure 1. The measurements at the Y K-edge (17300 eV) were made up to k = 12 Å^-1. A Lytle detector was used with helium gas flowing through the sample chamber and Ar gas through the detector chamber. The x-ray energy was calibrated with Y foil before the experiments. For harmonics rejection the monochromator was detuned by 50%. Because the doping concentration of yttrium is relatively low, fluorescence mode was used.

The fit results (Figure 2) indicate that the Y impurity does not exist as (Y¬_xTi)O_2-3x/2, i.e., the Y impurity is not structurally incorporated. This differs from the cases of doping TiO₂ with some other ions (e.g. Cr³⁺, Fe³⁺), where impurities are distributed within the nanoparticles [2,3]. The reason why Y³⁺ does not substitute for Ti⁴⁺ in TiO₂ nanoparticles is very likely the difference in the size of the Y³⁺ (0.892 Å to 1.10 Å) and Ti⁴⁺ ions (0.53 Å to 0.605 Å) [5]. Moreover, the first shell scattering in the EXAFS data is well fitted by Y-O bonds of length ~ 2.3 Å, a nearly 20% mismatch with the Ti-O bond lengths in the nanoparticles (1.92 Å and 1.96 Å) [6]. For heat-treated nanoparticles, the first shell Y-O coordination number is six, consistent with the yttrium oxide (Y₂O₃) structure. However, the characteristic XRD peaks of Y₂O₃ are not detected from our samples indicating that, if present, individual clusters of this phase should be smaller than ~1

Figure 2: Fourier transform magnitude of k3-weighted Y K-edge EXAFS spectra of yttrium-doped TiO2 nanoparticles as a function of treatment temperature. Inset: A representative one of the k3-weighted first-shell EXAFS spectra (solid green line) and the fit (dashed red line).